---
format: "aidr-story-markdown/v1"
id: "70b8ae71c9990d66079241ebcc14db12caa4ed933f23a523166766dae1102c5f"
canonical_url: "https://aidr.today/70b8ae71?lang=en"
title: "Meta Releases Muse Spark 1.3 Max for Strongest Reasoning Tier"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-09-04T20:52:10.000Z"
category: "Releases"
topics: ["meta","llm","coding","agent","reasoning"]
source_urls: ["https://huggingnews.com/ai/update-meta-releases-muse-spark-13-max-for-strongest-reasoning-tier-f2d45609","https://x.com/alexandr_wang/status/2095938990197329935","https://x.com/AIatMeta/status/2095940043294847084","https://x.com/alexandr_wang/status/2095939452002836524","https://x.com/finkd/status/2095960461879427299","https://x.com/AIatMeta/status/2095961487630389416"]
summary: "Developers can now access the max reasoning level of the Muse Spark 1.3 AI model through Muse Code and the Meta Model API. This variant provides stronger coding and agentic capabilities than the previously available \"high\" and \"xhigh\" versions of the 1.3 model. Users trigger the functionality by selecting Muse Spark 1.3 and adjusting the reasoning setting to \"max.\" The public rollout follows a limited preview for Meta partners and focuses on improving performance for complex coding tasks. Early user feedback describes the max reasoning variant as an efficient tool, as Meta seeks to provide frontier AI performance for developers building long-running, multi-agent workflows."
---

# Meta Releases Muse Spark 1\.3 Max for Strongest Reasoning Tier

> [Open the canonical story](<https://aidr.today/70b8ae71?lang=en>)

**Published:** 2026-09-04T20:52:10.000Z
**Category:** Releases
**Topics:** meta, llm, coding, agent, reasoning

## Summary

Developers can now access the max reasoning level of the Muse Spark 1\.3 AI model through Muse Code and the Meta Model API\. This variant provides stronger coding and agentic capabilities than the previously available "high" and "xhigh" versions of the 1\.3 model\. Users trigger the functionality by selecting Muse Spark 1\.3 and adjusting the reasoning setting to "max\." The public rollout follows a limited preview for Meta partners and focuses on improving performance for complex coding tasks\. Early user feedback describes the max reasoning variant as an efficient tool, as Meta seeks to provide frontier AI performance for developers building long\-running, multi\-agent workflows\.

## Sources

- [Story source](<https://huggingnews.com/ai/update-meta-releases-muse-spark-13-max-for-strongest-reasoning-tier-f2d45609>)
- [Story source](<https://x.com/alexandr_wang/status/2095938990197329935>)
- [Story source](<https://x.com/AIatMeta/status/2095940043294847084>)
- [Supporting source](<https://x.com/alexandr_wang/status/2095939452002836524>)
- [Supporting source](<https://x.com/finkd/status/2095960461879427299>)
- [Supporting source](<https://x.com/AIatMeta/status/2095961487630389416>)